| Name | Modified | Size | Downloads / Week |
|---|---|---|---|
| Parent folder | |||
| caprock_0.56.0_linux_amd64.tar.gz | 2026-09-09 | 8.7 MB | |
| caprock_0.56.0_windows_arm64.zip | 2026-09-09 | 8.2 MB | |
| checksums.txt | 2026-09-09 | 600 Bytes | |
| caprock_0.56.0_darwin_amd64.tar.gz | 2026-09-09 | 8.8 MB | |
| caprock_0.56.0_darwin_arm64.tar.gz | 2026-09-09 | 8.4 MB | |
| caprock_0.56.0_linux_arm64.tar.gz | 2026-09-09 | 8.1 MB | |
| caprock_0.56.0_windows_amd64.zip | 2026-09-09 | 8.9 MB | |
| README.md | 2026-09-09 | 1.6 kB | |
| v0.56.0 source code.tar.gz | 2026-09-09 | 7.3 MB | |
| v0.56.0 source code.zip | 2026-09-09 | 7.6 MB | |
| Totals: 10 Items | 65.9 MB | 0 | |
Added
- The context tax, in the screens you already look at. Every call re-sends
the whole conversation as a cache read before it does anything, and Caprock
stored every number needed to price that and showed none of them. It now
does, in three places and no new ones: the session card's Context caption
says what the next call costs at the current fill (
$0.07/call), the lifetime Breakdown carries aContext taxrow with its dollars and its share of what was priced, and a loop alert states what the repeated calls paid to re-read the conversation. On the author's own archive that row reads $11,117 of $13,910 — 80%.
Two limits are stated rather than papered over. Cache-read rates come from the pricing table per model, because Fable 5.1 and Mythos 5.1 read at 0.025x where the rest of the table reads at 0.1x. And a tool call is priced through the turn that paid for it, which the hook plane cannot identify: where some of a loop's calls are unlinked the alert says "at least", and the Breakdown row names the volume it had to exclude.
Fixed
- The recording stand outlived its terminal, forever. Eight of them were
found still running, the oldest two weeks old, each holding a daemon and a
temp directory open.
sleep 3600in a hold loop is a separate process, and bash will not run a trap until it returns — so a stand started detached never received the interrupt that would have stopped it. It waits on an interruptible sleep now. Affectsmake recordonly, not the daemon.